Physiologic and Metagenomic Attributes of the Rhodoliths Forming

Home , Bryozoa, Cnidaria, Crenarchaeota, Rhodolith

rhodoliths forming the largest CaCO3 bed in the South Atlantic Ocean

Giselle S Cavalcanti1, Gustavo B Gregoracci1, Eidy O dos Santos1, Cynthia B Silveira1, Pedro M Meirelles1, Leila Longo2, Kazuyoshi Gotoh3, Shota Nakamura3, Tetsuya Iida3, Tomoo Sawabe4, Carlos E Rezende5, Ronaldo B Francini-Filho6, Rodrigo L Moura1, Gilberto M Amado-Filho2 and Fabiano L Thompson1 1Institute of Biology, Federal University of Rio de Janeiro (UFRJ), Rio de Janeiro, Brazil; 2Instituto de Pesquisas Jardim Botaˆnico do Rio de Janeiro, Rio de Janeiro, Brazil; 3Laboratory of Genomic Research on Pathogenic Bacteria, International Research Center for Infectious Diseases, Research Institute for Microbial Diseases, Osaka University, Osaka, Japan; 4Laboratory of Microbiology, Hokkaido University, Sapporo, Japan; 5Environmental Science Laboratory, Campos dos Goytacazes, UENF, Rio de Janeiro, Brazil and 6Department of Engineering and Environment, Federal University of Paraı´ba, Paraı´ba, Brazil

Rhodoliths are free-living coralline algae (Rhodophyta, Corallinales) that are ecologically important for the functioning of marine environments. They form extensive beds distributed worldwide, providing a habitat and nursery for benthic organisms and space for fisheries, and are an important source of calcium carbonate. The Abrolhos Bank, off eastern Brazil, harbors the world’s largest 2 continuous rhodolith bed (of B21 000 km ) and has one of the largest marine CaCO3 deposits (producing 25 megatons of CaCO3 per year). Nevertheless, there is a lack of information about the microbial diversity, photosynthetic potential and ecological interactions within the rhodolith holobiont. Herein, we performed an ecophysiologic and metagenomic analysis of the Abrolhos rhodoliths to understand their microbial composition and functional components. Rhodoliths contained a specific microbiome that displayed a significant enrichment in aerobic ammonia- oxidizing betaproteobacteria and dissimilative sulfate-reducing deltaproteobacteria. We also observed a significant contribution of bacterial guilds (that is, photolithoautotrophs, anaerobic heterotrophs, sulfide oxidizers, anoxygenic phototrophs and methanogens) in the rhodolith metagenome, suggested to have important roles in biomineralization. The increased hits in aromatic compounds, fatty acid and secondary metabolism subsystems hint at an important chemically mediated interaction in which a functional job partition among eukaryal, archaeal and bacterial groups allows the rhodolith holobiont to thrive in the global ocean. High rates of photosynthesis were measured for Abrolhos rhodoliths (52.16 lmol carbon m 2 s 1), allowing the entire Abrolhos rhodolith bed to produce 5.65 105 tons C per day. This estimate illustrates the great importance of the Abrolhos rhodolith beds for dissolved carbon production in the South Atlantic Ocean. The ISME Journal (2014) 8, 52–62; doi:10.1038/ismej.2013.133; published online 29 August 2013 Subject Category: Microbe-microbe and microbe-host interactions Keywords: rhodoliths; holobionts; carbon cycle; biomineralization; Abrolhos Bank

Introduction The term holobiont refers to any organism and all of its associated symbiotic microbes (parasites, mutu- Eukaryote–microbe associations are increasingly alists, synergists and amensals) (Rosenberg and recognized as being essential for eukaryotic host Zilber-Rosenberg, 2011), including endobionts and health and metabolism, enabling the expansion of epibionts that perform diverse ecological roles their physiologic capacities (Rosenberg et al., 2007). (Wahl et al., 2012). A holobiont occupies and adapts to an ecological niche, and is able to employ Correspondence: FL Thompson, Av. Carlos Chagas Fo. S/N— strategies unavailable in any one species alone CCS—IB—BIOMAR—Lab de Microbiologia—BLOCO A (Anexo) when challenged by environmental perturbations. A3—sl 102, Cidade Universita´ria, Rio de Janeiro CEP 21941-599, The holobiont concept, first applied to corals after RJ, Brazil. E-mail: [email protected] they were found to host abundant and species- Received 25 March 2013; revised 23 June 2013; accepted 4 July specific microbial communities (Rohwer et al., 2013; published online 29 August 2013 2002; Reshef et al., 2006; Rosenberg et al., 2007), Rhodolith physiologic and metagenomic attributes GS Cavalcanti et al 53 has since been extended to the benthic alga (Barott ecological roles that rhodoliths have in the marine et al., 2011). These holobionts are diverse and realm. Rhodoliths may be key factors in a range of distinct from the microbiota in the surrounding invertebrate-recruitment processes, functioning as seawater and biofilms on abiotic surfaces (Barott autogenic ecosystem engineers by providing three- et al., 2011; Burke et al., 2011; Barott and Rohwer, dimensional habitat structures. A better understand- 2012). A wide range of beneficial and detrimental ing of the composition of and ecological interactions interactions have been described between macro- within the rhodolith holobiont would be a powerful algae and their microbiomes based on the exchange tool for the conservation and sustainable use of of nutrients, minerals and secondary metabolites these biological resources. This understanding (Hollants et al., 2012). could also provide insights into how cooperation Rhodoliths are biogenic calcareous structures and job partitioning between constituents contribute primarily formed by encrusting coralline algae to holobiont fitness, influence holobiont viability, (CCA; Corallinales, Rhodophyta). They provide and consequently affect associated ecosystems and highly biodiverse and heterogeneous habitats and the ubiquitous rhodolith worldwide distribution. are distributed worldwide (Foster, 2001; Steller We aimed to perform a metagenomic characteriza- et al., 2003; Nelson, 2009). Despite our vast knowl- tion of the Abrolhos rhodoliths to determine the edge concerning the ecophysiology and biogeogra- major taxonomic and functional components of phy of rhodoliths, some aspects of the biology of these organisms. We also examined the associated rhodoliths remain completely unknown. For fauna diversity and physiologic aspects (photosyn- instance, information on the microbial commu- thetic capacity and dissolved organic carbon (DOC) nities, the microbial metabolic strategies associated productivity) of the rhodoliths. with biogeochemical cycles and the biomineralization of CaCO3, and the photosynthetic productivity potential of rhodoliths is absent. Materials and methods Rhodolith beds constitute one of the Earth’s four Study site and sample collection major macrophyte-dominated benthic communities, This study was carried out in the Abrolhos Shelf off along with kelp beds, seagrass meadows and eastern Brazil. Rhodoliths were collected by scuba coralline algal reefs. Our group has mapped the diving in December 2010 from three different sites largest rhodolith bed in the world, which covers near two recently described sinkhole-like structures B 2 1 0 1 0 20 900 km of the Abrolhos Shelf (16 50 –19 45 S) called Buracas (Bastos et al., 2013). Buracas are cup- (Amado-Filho et al., 2012). The Abrolhos Bank is shaped depressions on the seafloor and their one of the largest marine CaCO3 deposits in the suggested function is to trap and accumulate organic world, with a production of 25 megatons of CaCO3 matter, thus functioning as productivity hotspots in per year (Amado-Filho et al., 2012). Nevertheless, the mid- and outer shelf of the central portion of the our limited knowledge of rhodolith biology hinders Abrolhos Bank (Cavalcanti et al., 2013). Seven our ability to develop a broader systemic under- rhodoliths were sampled as follows: two individual standing of their ecological role in the Abrolhos rhodoliths from the shallower portion (27 m) Bank. There are pressing concerns relating to the (17.813301 S/38.237441 W) outside the first Buraca; future of rhodolith beds as global carbon budgets, three from the inner region of the same Buraca (43 m and ocean acidification may interfere with rhodolith deep) (17.813991 S/38.243061 W) and two from a functioning and biogeochemical stability (Webster deeper point (51 m) (17.913611 S/37.909361 W) near et al., 2011; Whalan et al., 2012). In addition, another sinkhole-like structure (Supplementary because rhodoliths can be applied agriculturally to Table S1) away from the first (Figure 1). Mono- improve soil pH, rhodolith environments have specific rhodolith-forming CCA were selected by been suffering extensive commercial pressure. visual inspection. Immediately after collection, the In addition, rhodoliths are a non-renewable resource specimens were frozen in liquid nitrogen in the because of their extremely slow growth rates (Dias, field. For comparison purposes, surrounding water 2001; Barbera et al., 2003; Wilson et al., 2004). samples (8 l) collected using sterivex 0.2-mm filters Rhodoliths form biogenic matrices with complex at exactly the same points over the rhodolith beds at structures in which the interlocking branched thalli the first Buraca (inside—43 m; outside—27 m) can create microhabitats for diverse eukaryotic and outside the deeper Buraca (51 m) were used. assemblages, including epiphyte algae, microalgae A detailed description of the region of the Buracas, and different types of invertebrates from both hard water sampling and metagenomic characterization and soft benthos (Steller et al., 2003; Kamenos et al., of the planktonic microbial community is available 2004; Figueiredo et al., 2007; Riul et al., 2009; Bahia in the work by Cavalcanti et al. (2013). et al., 2010). Owing to the rich three-dimensional architecture, each rhodolith holobiont may be considered a small individual reef, providing a DNA extraction, pyrosequencing and sequence habitat for the young of various types of marine life analysis (for example, Arthropoda, Nematoda and Cnidaria). In the laboratory, a small fragment of each rhodolith Habitat and nursery functions are the most obvious sample (B1cm2) was sterilely macerated with liquid

The ISME Journal Rhodolith physiologic and metagenomic attributes GS Cavalcanti et al 54

Figure 1 Study site. The world’s largest rhodolith bed. (a) The Abrolhos bank, an expanse covering B46 000 km2 of the eastern Brazilian continental shelf where the rhodolith bed covers an area of B21 000 km2. Rhodoliths were sampled from two different sites surrounding the recently described sinkhole-like structures (red dots). (b) Physiognomy of Abrolhos rhodolith beds. (c) Individual monospecific rhodoliths.

nitrogen without the separation of epibionts to Benjamin–Hochberg FDR multiple test correction. provide a representation of the entire rhodolith A principal component analysis was conducted holobiont as previously defined (calcifying algae using the STAMP software package to compare the plus associated microbes, fauna and flora). taxonomic grouping based on the taxonomic class A subsequent step using CTAB buffer with 100 mM contributions of each metagenome from both the of EDTA and a PowerSoil purification column was rhodolith and the surrounding seawater. To isolate used to gather the DNA of high-molecular-weight the relative contributions of taxonomic groups rhodolith holobionts as described by Garcia et al. within the rhodolith metagenomes, the most (2013). High-quality total DNA extracted from abundant groups (Alphaproteobacteria, Gammapro- rhodoliths was then sequenced using a 454 teobacteria, Eukarya, Betaproteobacteria, Deltapro- GS Jr machine (454 Life Sciences, Branford, CT, teobacteria, Actinobacteria, Firmicutes, Bacteroidetes, USA) and the GS Jr Titanium-sequencing process. Cyanobacteria, Planctomycetes, Archaea and Others) Sequences were submitted to the MG-RAST 3.1 were functionally re-annotated using the Work- server (Metagenomics-Rapid Annotation Using bench tool, and their functions were compared with Subsystems Technology) (Meyer et al., 2008) and an analysis of variance using a Tukey–Kramer post- quality filtered. Post-quality-control (QC) sequences hoc test, an eta-squared effect size and a Benjamin– were annotated using the (SEED) Subsystems Hochberg multiple test correction. In all these cases, Technology for functional classification (Overbeek P-values of o5% were considered statistically et al., 2005) and the GenBank database for phyloge- significant. netic analyses. All BLAST queries were performed with a maximum expected cutoff value of 10–5. Associated fauna analysis For each sampling point, the fauna associated with the rhodolith was analyzed. These rhodoliths were Statistical analyses preserved in 10% formalin, and the faunal organ- The Statistical Analysis of Metagenomic Profiles isms were analyzed with regard to the high-level (STAMP v.2.0.0) software was used for statistical taxonomic groups, such as phylum, class, order and analysis (Parks and Beiko, 2010). For comparison family, when possible using stereomicroscopy. purposes, rhodolith metagenomes were compared with water metagenomes from the same site (Cavalcanti et al., 2013). The water and rhodolith Primary productivity of rhodoliths samples were compared by using a two-sided To determine the photosynthetic activity of rhodo- Welch’s t-test with 95% confidence intervals calcu- liths in the Abrolhos Bank, we applied the lated by inverting the Welch’s test and by using the pulse-amplitude-modulated method (PAM) that

The ISME Journal Rhodolith physiologic and metagenomic attributes GS Cavalcanti et al 55 uses variations in the effective quantum yield of controls were performed in three replicates, and photosystem II as a method to assess the photosyn- the experiment was repeated twice. Incubation was thetic performance of photosystem II (Schereiber carried out in situ (at a depth of 30 m) for 2 h from et al., 1966). Briefly, the change in fluorescence 1300 to 1500 hours using water from near the bottom emitted by chlorophyll before and after the stimula- of the seafloor—the same water mass to which tion with a saturating light pulse is used to calculate rhodoliths are normally exposed. Dissolved organic the effective quantum yield (DF/Fm0) (Maxwell and carbon (DOC) was measured at time 0 and 2 h later, Johnson, 2000). The underwater DIVING PAM and the variation in DOC concentrations was (Walz, Germany) was used directly in the field with analyzed. For the DOC measurements, bottled water a red light under ambient light. DIVING PAM samples were filtered using pre-combusted (550 1C) measurements took place 33 m deep in rhodolith Whatman GF/Filters, preserved using 10% H3PO4, beds at 1300 hours on 17 August 2011. A total of 58 and stored and refrigerated in amber glass bottles. measurements were taken in 12 different rhodolith Samples were further acidified with 2 N HCl and specimens—five measurements per specimen. sparged with ultrapure air, and the DOC was A consistent distance of 1 cm between the fiberoptic determined by high-temperature catalytic oxidation sensor and the rhodolith surface was maintained in a Shimadzu TOC 5000 Analyzer (Ovalle et al., using an adaptation of the DIVING PAM sample 1999). holder. The yield (DF/Fm0) was measured using a 0.8 period of saturating light (ca. 8000 mmol m 2 per s), and an absorption factor of 0.56 was applied for Results electron transport rate (ETR) calculations, as previously determined for Rhodophyta (Beer and A total of 60.85 million non-redundant nucleotide Axelsson, 2004) (ETR ¼ DF/Fm0 PAR 0.5 0.56, bases (Mb) were generated in this study. Approxi- where PAR is the photosynthetic active radiation at mately 165 000 high-quality metagenome sequences the sampling site and 0.5 is the factor that accounts from seven different rhodoliths from three different for the distribution of electrons between photosys- sampling sites in the outer region of the Abrolhos tem I and photosystem II). Similarly, the theoretical Bank were obtained. After quality control, each rhodolith metagenome contained between 17 863 model of 0.25 mol of O2 for each mol of electrons was also used in the ETR calculations (Beer and and 32 488 reads, in which an average of 24% was Axelsson, 2004). classified by MG-RAST taxonomically (a total of 41 051 sequences) and functionally (a total of 39 691 sequences) (Supplementary Table S1). The major contributing domain was Bacteria (B83.64%). The Rhodolith dissolved organic carbon release B Incubation experiments were performed to deter- contribution of Archaea was 3%, whereas that of mine the amount of fixed carbon released by the the Eukarya domain was the most variable, ranging rhodoliths in the form of exudates. Single rhodo- from 6.5 to 27.4% (Supplementary Table S1). liths, B10 cm in diameter, were randomly collected and placed inside translucent Niskin bottles, which Taxonomic assignment of rhodolith metagenomes allow photosynthetically active radiation to pass Rhodolith replicates harbor impressive homogenous through for light treatment, or inside covered Niskin and characteristic communities of Bacteria (Figure 2). Bottles (dark) as a control. The treatments and When compared with the surrounding seawater

Figure 2 Homogeneity of the rhodolith metagenome. Rhodoliths and the surrounding seawater from depths of 27, 43 and 51 m were compared. (a) Principal component analysis (PCA) grouping rhodolith (red) and seawater (blue) metagenomes by class contribution. (b) Taxonomic composition of the rhodolith and surrounding water metagenomes. A taxonomic assignment was performed using MG-Rast based on sequence similarities to the GenBank database.

The ISME Journal Rhodolith physiologic and metagenomic attributes GS Cavalcanti et al 56 metagenomes, rhodolith metagenomes from different the variability in the surrounding water detailed and sampling sites clustered closely together based on discussed by Cavalcanti et al. (2013) (Figure 2b). The class-contribution grouping, despite the clustering of majority of bacterial sequences were assigned to the surrounding water samples based on site and depth Proteobacteria phylum (an average of 47.13%), (Figure 2a). The bacterial class composition was followed by Actinobacteria, Firmicutes and Cyano- extremely consistent among the seven rhodoliths and bacteria, contributing B5% each (Figure 3a). did not show any site or depth correlations, despite The Alpha- and Gammaproteobacteria together were

Figure 3 Taxonomic composition of rhodolith metagenomes. A taxonomic assignment was performed using MG-Rast based on sequence similarities to the GenBank database. (a) The average percentage of sequences from rhodolith and the surrounding water metagenome with the best BLAST similarity to Archaeal and Bacterial domains (phylum Proteobacteria and other Bacteria phyla). (b) More detailed breakdown of the Proteobacteria group. Relative contribution of the main proteobacterial orders in the rhodolith and the surrounding seawater metagenomes. Comparison of taxonomic contributions between rhodoliths and the surrounding water: (*) represents groups that are significantly over-represented in the rhodolith metagenome (Po0.5); (¤) represents groups that are significantly over-represented in the seawater metagenome (Po0.5).

The ISME Journal Rhodolith physiologic and metagenomic attributes GS Cavalcanti et al 57 responsible for 72.65% of the Proteobacterial below, and the rest are summarized in sequences; nevertheless, a significantly greater con- Supplementary Table 3. Briefly, fermentation tribution was observed for Beta- and Deltaproteo- (P ¼ 1.92 10–2)—particularly acetone butanol etha- bacterial classes relative to the surrounding water. nol (ABE) synthesis (P ¼ 2.21 10–2) and butanol This increased contribution was observed for the biosynthesis (P ¼ 2.09 10–2), sulfatases and mod- majority of orders of Beta- and Deltaproteobacteria, ifying factors (P ¼ 8.31 10–4), fatty acids with a remarkable contribution for Burkolderiales (P ¼ 6.24 10–3), ABC transporters (P ¼ 3.74 10–4), (over 5%) (Figure 3b). The other rhodolith-holobiont metabolism of central aromatic intermediates components, Archaea and Eukarya, had a more (P ¼ 3.27 10–2), peripheral pathways for the variable constitution between individual rhodoliths. catabolism of aromatic compounds (P ¼ 3.37 10–2), The most abundant archaeal phylum was Thaumar- biologically active compounds in metazoan chaeota (ranging from 0.48 to 4.12%), with a greater cell defense (P ¼ 2.76 10–2), steroid sulfates contribution in the samples obtained from deeper (P ¼ 2.06 10–2), galactosylceramide and sulfatide water (51 m). In addition to Thaumarchaeota, both metabolism (P ¼ 3.05 10–3), and organic sulfur assim- Euryarchaeota and Crenarchaeota phyla (a total ilation (P ¼ 3.36 10–2)—was increased in the rhodo- of 1.26%) were statistically more abundant in liths. Alternatively, DNA replication (P ¼ 2.73 10–3), the rhodolith metagenomes compared with the the recBCD DNA repair pathway (P ¼ 3.97 10–2), surrounding water (Figure 3a). For the Eukaryota bacterial cell division (P ¼ 2.84 10–2), the organic- domain—the most variable domain in our study— acid-generating methylcitrate cycle (P ¼ 3.41 10–3), over 60% of all the sequences were identified as and the propionate-CoA to succinate module invertebrate marine phyla Chordata, Arthropoda (P ¼ 9.34 10–3), were reduced in the rhodoliths and Cnidaria, and as algal groups Streptophyta compared with the surrounding water. The comparisons and Chlorophyta (Supplementary Table S2). of the most abundant groups revealed distinguishing features. Most metabolic pathways were shared between all the groups, with the following note- Metabolic diversity of rhodolith metagenomes worthy exceptions. Potassium metabolism was The overall functional annotation (subsystems level 1) higher in Alphaproteobacteria (P ¼ 1.15 10–5). of rhodoliths was compared with the surroun- Respiration (P ¼ 2.77 10–3) and photosynthesis ding water metagenomes (Figure 4). Rhodoliths (P ¼ 8.46 10–7) were increased in Eukarya, with presented more sequences related to fatty acids, reasonably large representation in Archaea for lipids and isoprenoids (P ¼ 4.38 10–2), metabolism respiration and in Cyanobacteria for photosynthesis. of aromatic compounds (P ¼ 7.9 10–3), secondary Proteolytic pathways were also more abundant in metabolism (P ¼ 4.88 10–4) and sulfur metabolism Eukarya (P ¼ 9.64 10–4). Iron acquisition and meta- (P ¼ 6.97 10–3). The water samples had more bolism were more prominent in Bacteroidetes sequences related to cell division and cell (P ¼ 7.53 10–3), although Gammaproteobacteria cycle (P ¼ 3.76 10–2), iron acquisition and metabo- were enriched in siderophores (P ¼ 5.57 10–3), lism (P ¼ 3.31 10–2), and RNA metabolism and Alphaproteobacteria had an over-representation (P ¼ 3.47 10–2). At the second and third levels, of heme and hemin uptake hits (P ¼ 1.6 10–2). other observed noteworthy differences are listed Archaea showed a predominant representation

Figure 4 Functional profile. Relative contribution of subsystems (first-level hierarchy) in the rhodolith and surrounding seawater metagenomes. Comparison of functional contributions between rhodoliths and the surrounding water: (*) represents subsystems that are significantly over-represented in the rhodolith metagenome (Po0.5); (¤) represents subsystems that are significantly over-represented in the seawater metagenome (Po0.5).

The ISME Journal Rhodolith physiologic and metagenomic attributes GS Cavalcanti et al 58 of several subsystems, including organic acids On the sample from the depth of 51 m, a morphotype (P ¼ 3.3 10–2), adhesion (P ¼ 1.86 10–5), denitrifi- of Brachiopoda and residuals of Echinoidea (Echi- cation (P ¼ 9.91 10–3) and inorganic sulfur assim- nodermata) calcareous skeletons were found. The ilation (P ¼ 8.71 10–7), although none of these roles sample from a depth of 27 m presented a different were exclusive to this domain. Betaproteobacteria fauna composition from the others, with a large were distinguished by fatty-acid degradation rhodolith surface area covered by colonial forms of (P ¼ 7.42 10–3), and Deltaproteobacteria were Ascidiacea (phylum Chordata, subphylum Tuni- distinguished by the invasion and intracellular cata)andPorifera, in addition to Bryozoa, Forami- resistance subsystem (P ¼ 1.3 10–2). Some groups niferea, Polychaeta and Vermetidae, completing the (Actinobacteria, Bacteroidetes, Cyanobacteria and epifauna. The most abundant groups, primarily Firmicutes) had higher numbers of DNA repair hits because of their encrusting colonial habits, were (P ¼ 3.01 10–4), but this subsystem was well repre- Ascidiacea and Porifera, followed by Bryozoa sented in most bacterial groups. The cAMP signaling and Foraminiferea. In this sample, we observed was highly represented in Cyanobacteria and Beta- the occurrence of other groups, including the proteobacteria (P ¼ 1.75 10–5). Type III protein phyla Platyhelminthes, Nematoda and Arthropoda secretion system orphans (P ¼ 4.9 10–2), type IV (subphylum Crustacea), and the class Polyplaco- secretion systems (P ¼ 9.16 10–3) and type VIII phora (phylum Mollusca), although with only one secretion systems (P ¼ 4.4 10–2) were also specimen registered for each group. increased in Gammaproteobacteria, Betaproteo- bacteria and Bacteroidetes, respectively. Primary productivity of rhodoliths Photosynthetically active radiation from Litho- Fauna taxonomic identification thamnion crispatum at a depth of 33 m, the site of The most frequent taxonomic groups associated measurements, was 1436±47.84 mmol photons m 2s 1 with the sampled rhodoliths were the Foramini- (average±s.e.). The effective quantum yield of ferea, Bryozoa, Annelida and Mollusca phyla, which photosystem II determined for the rhodoliths ranged were observed in all the samples, although repre- from 0.206 to 0.78, with a mean value of 0.511, sented by different morphotypes. Regarding the which allowed linear correlations between ETR and epifauna, colonial encrusting forms covered large CO2 fixation. These values were considerably above rhodolith areas, where the most abundant group was the 0.1 critical limit value for these correlations. Bryozoa, followed by Foraminiferea, particularly The ETR was 208.67±10.19 mmol electrons m 2 s 1

Homotrema rubrum. The phylum Annelida was (average±s.e.). Therefore, the calculated CO2 fixa- largely represented by the encrusting calcareous tion rate was 52.16±2.57 mmol carbon m 2 s 1 or tubes of Polychaeta. The occurrence of the 27.03 g carbon m 2 per day (Table 1). phylum Mollusca was observed by the residual shells from Bivalvia and Gastropoda, particularly from the Vermetidae family. The infauna was mainly Rhodolith organic carbon release composed of calcareous tubes of Polychaeta, The concentration of DOC in the water just above followed by Bryozoa and Foraminiferea. Different the rhodolith beds was 1.85±0.32 mg l 1 (avera- morphotypes (approximately five) of free-living ge±s.d.). The DOC concentration increased after the Foraminiferea were observed in all the samples, 2 h of incubation, with DDOC of 0.13 and 0.04 mg l 1 showing the highest species richness compared with for light and dark bottles, respectively. The control the other groups of fauna associated with rhodoliths. bottles had DOC consumptions of 0.26 and Rhodoliths from the depths of 43 and 51 m 0.28 mg l 1 for light and dark bottles, respectively, showed lower species richness compared with the likely because of the planktonic activity in the 27 m sample. Rhodoliths from a depth of 43 m were Niskin bottles. Discounting the amount of organic mainly associated with filamentous green algae, and carbon consumed by the plankton, we observed that the sample from this depth was the only containing rhodoliths released 0.39±12 and 0.32±19 mg l 1 of one specimen and shell marks of Lithophaga (Bivalvia). DOC in light and dark incubations, respectively,

Table 1 Calcifying algae productivity. Photosynthetic capacity of free-living rhodolith holobionts and crustose coralline algae

Organism Depth Temperature Irradiance Fixed carbon Reference (m) (1C) (mMol photons m 2 s 1) (mMol C m 2 s 1)

Rhodoliths 33 26 1436±47 52.1±2.5 This study CCA 6–7 26 9207±907 274.5±36.5 This study P. foecundum (CCA) 15–18 0–1 o5 45.2–66.9 Roberts et al., 2002 P. tenue (CCA) 15–18 0–1 o5 42.5–47.2 Roberts et al., 2002 M. Engerlhartii (CCA) 16–20 0–1 o5 9.4–17.9 Schwarts et al., 2005

The ISME Journal Rhodolith physiologic and metagenomic attributes GS Cavalcanti et al 59 showing no significant difference between these anoxygenic phototrophs (that is, purple (2.33%) and two treatments (P ¼ 0.8). The release rate of organic green sulfur bacteria (1.33%)) and methanogens carbon was 0.04 mg ± 0.01 DOC h 1dm 2 of rhodo- (0.46%) in the rhodolith metagenome. These guilds lith surface area, or 3.58 ± 1.46 mmol h 1 dm 2. with similar metabolic roles operate in collaboration to accomplish the cycling of key elements (O, N, S and C), which may allow them to carry out Discussion important tasks, such as biomineralization (Dupraz et al., 2009). Microbially induced carbonate pre- Our rhodolith metagenomic analysis demonstrated cipitation, which occurs in microbialites, is carried the wealth of microorganisms that inhabit this out by the interaction of microbial consortia, their holobiont. We verified an extremely similar compo- growth and metabolism, cell surface properties, and sition among the microbiomes of the rhodoliths from extracellular polymeric substances (Dupraz et al., the three different sampling sites in the Abrolhos 2009). Calcium carbonate formation and precipita- Bank—a possible reflection of a long-term host– tion are central processes in rhodolith construction. microbe association. This association may enable The corallinaceae algae are considered to be major the holobiont to expand its physiological capacity factors in mineralization in rhodoliths. Compara- and geographic distribution (Hollants et al., 2012). tively little attention has been paid to the role of Polychaetes, crustaceans (amphipods), ophiuroids rhodolith microbes in calcium carbonate formation and mollusks, commonly found in the rhodolith and precipitation. Although the diversity and infauna of the Abrolhos Bank, were also the main abundance of microbes provide only a cursory invertebrate taxa inhabiting the rhodoliths analyzed indication of what is metabolically feasible, it is in this study. Our metagenomic analysis extended noteworthy that in the rhodolith microbiome, these previous observations (Figueiredo et al., 2007), we found major functional groups related to micro- detecting other types of flora and fauna (for bial-induced organomineralization, which may example, foraminifers). suggest an important but still unappreciated role of It is plausible that the genesis and growth of microbes in carbonate precipitation in rhodoliths. rhodoliths are intimately related to their microbial This functional metagenomic profile offers insights constituents. The remarkable homogeneity revealed into the metabolism of the rhodolith holobiont. in the rhodolith microbiomes hints at the existence Whereas water-column microbes seem more prone of a close relationship between the microbes to cell division, DNA repair and iron acquisition, and their rhodolith host. Indeed, despite a lack of rhodolith-associated microbes are more specialized knowledge of the principles underlying the assem- in chemical communication. This interplay is bly and the structure of these complex microbial suggested by the increased representation of secondary communities, other examples of host specificity metabolites, metazoan-active cell compounds and have also been illustrated for macroalgal–bacterial specialized bacterial secretion systems in the rhodo- interactions (Hollants et al., 2012). Most of the lith metagenome. Fatty-acid metabolism and degra- rhodolith metagenomic sequences (over 70%) were dation are also important, hinting that an interaction identified as bacterial protein-coding sequences. likely occurs in the algal cell membrane. A notable Some bacterial taxa, such as Alphaproteobacteria, amount of capsular and extracellular polysaccharide Firmicutes and Actinobacteria, are as abundant in hits indicate a structured biofilm in association with the rhodolith holobiont (19.7, 5.8 and 5.8%, respec- this surface. Although the rhodolith biofilm is clearly tively) as in other red seaweed–microbe holobionts less impressive than coral mucus or microbial mats, (13, 10 and 9%, respectively). However, some taxa the presence of certain taxonomic groups suggests were particularly enriched in rhodoliths (for example, that a considerable degree of spatial structuring Betaproteobacteria with 6.5% and Deltaproteo- occurs in this biofilm, providing microenvironments bacteria with 5.7%) compared with other red that support different metabolic guilds, akin to corals seaweeds (Hollants et al., 2012). The metabolically (Garcia et al., 2013), and microbial mats (Dupraz diverse groups of the aerobic ammonia-oxidizing et al., 2009). Interestingly, it has long been known betaproteobacteria Nitrosomonadales, Rodhocy- that the cell wall or mucilage is the calcification site clales and Hydrogenophylales, and the dissimilative in coralline red algae, such as rhodoliths (Borowitzka, sulfate-reducing deltaproteobacteria Desulfovibrio- 1987), perhaps with an intravesicular stage that could nales, Desulfobacterales and Desulforomonadales involve the Golgi apparatus (Krumbein and Larkum, were significantly increased in the rhodolith micro- 1987). Mineralization in these algae is strictly biome compared with the surrounding water. controlled, as every aspect of the process is regulated by the eukaryotic cell (Borowitzka, 1987; Dupraz et al., 2009). Given this close affinity, it is likely that Job partitioning in the rhodolith holobiont the microbial composition and metabolic rates asso- We observed significant contributions of photo- ciated with this structure will influence its physico- lithoautotrophs (that is, cyanobacteria (5.09%)), anae- chemical characteristics (Dupraz and Visscher, 2005; robic heterotrophs (predominantly sulfate-reducing Dupraz et al., 2009). The promotion ( þ ) or hindrance proteobacteria (1.52%)), sulfide oxidizers (0.37%), ( ) of calcium carbonate precipitation is

The ISME Journal Rhodolith physiologic and metagenomic attributes GS Cavalcanti et al 60 determined by the net effect of competing and potassium cycling and iron inputs together with antagonistic microbial metabolic processes, which Bacteroidetes and Gammaproteobacteria. Archaea cannot be predicted based only on the functions input inorganic sulphur into the system but revealed by metagenomics (Khodadad and Foster, secrete organic acids, possibly hindering carbonate 2012). However, our study revealed interesting deposition. features of the rhodolith metagenome. For instance, the main energy production and carbon fixation pathways appear to be via algal photosynthesis ( þ ). Rhodoliths are the major dissolved carbon biofactories Cyanobacteria appear to have a secondary role in Notably, the rhodolith holobiont photosynthetic this system. The main carbon-cycling balance capacity measured in this study via a PAM-based (photosynthesis ( þ )/respiration( )) is performed analysis showed high photosynthesis rates for by Eukarya, unlike in stromatolites (Dupraz et al., Abrolhos rhodoliths (52.16 mmol carbon m 2 s 1) 2009). Microbial metabolism may still have a role in when compared with other crustose coralline algae, terms of anoxygenic photosynthesis ( þ ), fermenta- exemplified by Phymatolithon foecundum (45.2–66.9), tion ( ), and aerobic ( ) and anaerobic respirations Phymatolithon tenue (42.5–47.2) and Mesophyllum ( þ ) (sulfate reduction, denitrification) (Krumbein engelhartii (9.4–17.9) (Table 1) (Roberts et al., 2002; and Larkum, 1987; Visscher et al., 1998; Dupraz and Schwarz et al., 2005). Nevertheless, the gross Visscher, 2005; Dupraz et al., 2009). Although some photosynthetic rate of 27.03 g C m 2 per day calcu- organic-acid-generating metabolic pathways ( ) are lated for the rhodolith beds was far below the reduced in rhodoliths, the higher representation of previous values of 462 and 2062 mg C m 2 per day fermentation ( ), butanol biosynthesis ( ) and for coastal and open ocean mesotrophic phytoplankton, ABE synthesis ( ) suggest a negative role for respectively (Liu et al., 1997; Mora´n et al., 2002). microbes in rhodolith calcification. However, sulfate These data may suggest that the rhodoliths also oxidizers ( þ ) and anoxygenic photosynthesis ( þ ) function as a heterotrophic system because of their guilds were detected in this holobiont and could large associated fauna, but that part of the fixed significantly promote the precipitation of calcium carbon is released as DOC. We calculated a rate of carbonate in the rhodoliths. 3.58 mmol C h dm 2 organic carbon released in the Another potential influence of the microbes in same range found for crustose coralline algae in rhodoliths is in sulfur cycling, similar to modern Moorea, French Polynesia. On the basis of the marine stromatolites in the Bahamas (Visscher et al., 20 902 km2 of bank extension determined previously 1998) and Mexico (Breitbart et al., 2009). The (Amado-Filho et al., 2012) and the photosynthetic negatively charged sulfated polysaccharides of the rates calculated here, the estimated total photosyn- algal cell wall are likely the ligands for Ca2 þ ions thetic activity for the Abrolhos Bank is calculated to (Borowitzka, 1987). Theoretically, these polysac- be 56.5 104 tons C per day. This estimate may charides can also limit carbonate precipitation by illustrate the great importance of the Abrolhos sequestering these ions in the organic matrix rhodolith beds for DOC production in the South (Dupraz et al., 2009). We also observed steroid Atlantic Ocean. sulfates. The presence of sulfatases, sulfatide metabolism and organic sulfur assimilation pathways indicates a microbial role in the cleavage and Conclusion cycling of these sulfur radicals, similar to the model proposed for microbialites (Breitbart et al., 2009). The present study extends the previous studies on The importance of degradation of fatty acids and rhodoliths (Amado-Filho et al., 2012), exploring the several aromatic compounds through central and possible roles of microbes in the structure and peripheral pathways also suggests a degradation of function of rhodoliths, as well as the systemic roles the mucilage/biofilm interface. The release of the of rhodoliths in the production of DOC. Calci- bound Ca2 þ ions and carboxyl groups through fying algae have a significant role in the ocean degradation by aerobic respiration (Breitbart et al., carbon cycle, not only in the deposition of 2009) or consortia between sulfate reducers and calcium carbonate but also in their photosynthetic fermenters (Dupraz et al., 2009) can also promote capacity. Future studies utilizing metatranscrip- mineralization independently of the light/oxygen tomics (Khodadad and Foster, 2012), stable isotopes conditions, which could explain the low rates of (Breitbart et al., 2009) or microelectrodes (Visscher dark calcification sometimes observed (Borowitzka, et al., 1998) could extend these observations and 1987; Krumbein and Larkum, 1987). This accessory shed new light on the biology of rhodoliths. Our role of precipitation should be most prevalent in the first attempt to characterize the microbiome of the space between algal cells and between these cells rhodoliths allowed us to determine the major and the substratum, in which the influence of the microbial components of the rhodolith holobiont. coralline algae would be lessened. This collective The impressively homogeneous bacterial contribu- role would explain the selective nature of some tion among rhodolith replicates, coupled with the of the microbes associated with this holobiont. increased representation of pathways associated Alphaproteobacteria could also be responsible for with aromatic compounds, fatty acids and

The ISME Journal Rhodolith physiologic and metagenomic attributes GS Cavalcanti et al 61 secondary metabolism, suggests an important che- Burke C, Thomas T, Lewis M, Steinberg P, Kjelleberg S. mically mediated interaction within the rhodolith (2011). Composition, uniqueness and variability of the holobiont. We propose that the rhodolith microbial epiphytic bacterial community of the green alga Ulva guilds, based on their extended metabolic capacity, australis. ISME J 5: 590–600. have important roles in the rhodolith holobiont Cavalcanti GS, Gregoracci GB, Longo LL, Bastos AC, functioning—for example, in biomineralization. Ferreira CM, Francini-Filho RB et al. (2013). Sinkhole- like structures as bioproductivity hotspots in the Abrolhos Bank. Continental Shelf Research (in press). Dias GTM. (2001). Granulados biocla´sticos—algas calca´r- Conflict of Interest ias. BrazJ Geophys 18: 307–308. Dupraz C, Visscher PT. (2005). Microbial lithification in The authors declare no conflict of interest. marine stromatolites and hypersaline mats. Trens Microbiol 13: 429–438. Dupraz C, Reid RP, Braissant O, Decho AW, Norman RS, Acknowledgements Visscher PT. (2009). Processes of carbonate precipitation in modern microbial mats. Earth-Sci Rev 96: 141–162. We acknowledge SISBIOTA CNPq/FAPES, FAPERJ, CNPq Figueiredo MADO, Menezes KSD, Costa-paiva EM, Paiva PC, and CAPES for funding. We also acknowledge the Project Ventura CRR. (2007). Experimental evaluation of for the International Research Center for Infectious rhodoliths as living substrata for infauna at the Diseases, Research Institute for Microbial Diseases, Osaka Abrolhos Bank, Brazil. Cienc Mar 33: 427–440. University, from the Ministry of Education, Science, Foster MS. (2001). Rodotlihs: between rocks and soft Sports, Culture, and Technology, Japan. FT. Giselle places. J Phycol 37: 659–667. Cavalcanti would like to thank the CNPq for research Garcia GD, Gregoracci GB, de OSE, Meirelles PM, Silva GG, and PhD fellowships. This paper is part of the DSc Edwards R et al. (2013). Metagenomic analysis requirements for Giselle da Silva Cavalcanti in the of healthy and white plague-affected Mussismilia Biodiversity and Evolutionary Biology Graduate Program braziliensis Corals. Microb Ecol 65: 1076–1086. of the Federal University of Rio de Janeiro. Hollants J, Leliaert F, De Clerck O, Willems A. (2012). What we can learn from sushi: a review on seaweed- bacterial associations. Fems Microbiol Ecol 83: 1–16. Kamenos NA, Moore PG, Hall-Spencer JM. (2004). Maerl grounds provide both refuge and high growth poten- References tial for juvenile queen scallops (Aequipecten opercu- laris L.). Mar Ecol 313: 241–254. Amado-Filho GM, Moura RL, Bastos AC, Salgado LT, Khodadad CLM, Foster JS. (2012). Metagenomic and Sumida PY, Guth AZ et al. (2012). Rhodolith beds are metabolic profiling of nonlithifying and lithifying major CaCO3 bio-factories in the tropical South West stromatolitic mats of Highborne Cay, The Bahamas. Atlantic. PLoS One 7: e35171. PLoS One 7. Bahia RG, Abrantes DP, Brasileiro PS, Pereira GH, Krumbein MA, Larkum AWD. (1987). Calcification in Amado GM. (2010). Rhodolith bed structure along a algae: mechanisms and the role of metabolism. Crit depth gradient on the northern coast of Bahia state, Rev Plant Sci 6: 1–45. Brazil. Braz J Oceanogr 58: 323–337. Liu H, Nolla HA, Campbell L. (1997). Prochlorococcus Barbera C, Bordehore C, Borg JA, Glemarec M, Grall J, growth rate and contribution to primary production in Hall-Spencer JM et al. (2003). Conservation and the equatorial and subtropical North Pacific Ocean. management of northeast Atlantic and Mediterranean Aquat Microb Ecol 12: 39–49. maerl beds. Aquat Conserv 13: S65–S76. Maxwell K, Johnson GN. (2000). Chlorophyll fluorescence—a Barott KL, Rodriguez-Brito B, Janousˇkovec J, Marhaver KL, practical guide. JExpBot51: 659–668. Smith JE, Keeling P et al. (2011). Microbial diversity Meyer F, Paarmann D, D’Souza M, Olson R, Glass EM, associated with four functional groups of benthic Kubal M et al. (2008). The metagenomics RAST reef algae and the reef-building coral Montastraea server—a public resource for the automatic phyloge- annularis. Environ Microbiol 13: 1192–1204. netic and functional analysis of metagenomes. BMC Barott KL, Rohwer FL. (2012). Unseen players shape Bioinformatics 9: 386. benthic competition on coral reefs. Trends Microbiol Mora´n XAG, Estrada M, Gasol JM, Pedro´s-Alio´ C. (2002). 20: 621–628. Dissolved primary production and the strength of Bastos AC, Moura RL, Amado-Filho GM, D’Agostini DP, phytoplankton- bacterioplankton coupling in contrast- Secchin NA, Francini-Filho RB et al. (2013). Buracas: ing marine regions. Microbial Ecol 44: 217–223. novel and unusual sinkhole-like features in the Nelson WA. (2009). Calcified macroalgae—critical to Abrolhos Bank. Cont Shelf Res (in press). coastal ecosystems and vulnerable to change: a review. Beer S, Axelsson L. (2004). Limitations in the use of PAM Mar Freshwater Res 60: 787–801. fluorometry for measuring photosynthetic rates of Ovalle ARC, Rezende CE, Carvalho CEV, Jennerjahn TC, macroalgae at high irradiances. Eur J Phycol 39: 1–7. Ittekkot V. (1999). Biogeochemical characteristics Borowitzka MA. (1987). Calcification in Algae—mechanisms of coastal waters adjacent to small river-mangrove and the role of metabolism. Crit Rev Plant Sci 6: 1–45. systems, East Brazil. Geo-Mar Lett 19: 179–185. Breitbart M, Hoare A, Nitti A, Siefert J, Haynes M, Overbeek R, Begley T, Butler RM, Choudhuri JV, Chuang HY, Dinsdale E et al. (2009). Metagenomic and stable Cohoon M et al. (2005). The subsystems approach to isotopic analyses of modern freshwater microbialites genome annotation and its use in the project in Cuatro Cienegas, Mexico. Environ Microbiol 11: to annotate 1000 genomes. Nucleic Acids Res 33: 16–34. 5691–5702.

The ISME Journal Rhodolith physiologic and metagenomic attributes GS Cavalcanti et al 62 Parks DH, Beiko RG. (2010). Identifying biologically of non-geniculate coralline algae at Cape Evans, relevant differences between metagenomic communities Ross Sea, Antarctica. Mar Ecol Prog Ser 294: Bioinformatics 26: 715–721. 131–140. Reshef L, Koren O, Loya Y, Zilber-Rosenberg I, Rosenberg E. Steller DL, Riosmena-Rodriiguez R, Foster MS, Roberts CA. (2006). The coral probiotic hypothesis. Environ (2003). Rhodolith bed diversity in the Gulf Microbiol 8: 2068–2073. of California: the importance of rhodolith structure Riul P, Lacouth P, Pagliosa PR, Christoffersen ML, and consequences of disturbance. Aquat Conserv 13: Horta PA. (2009). Rhodolith beds at the easternmost S5–S20. extreme of South America: community structure of an Visscher PT, Reid RP, Bebout BM, Hoeft SE, Macintyre IG, endangered environment. Aquat Bot 90: 315–320. Thompson JA. (1998). Formation of lithified Roberts RD, Kuhl M, Glud RN, Rysgaard S. (2002). Primary micritic laminae in modern marine stromatolites production of crustose coralline red algae in a high (Bahamas): The role of sulfur cycling. Am Mineral Arctic fjord. J Phycol 38: 273–283. 83: 1482–1493. Rohwer F, Seguritan V, Azam F, Knowlton N. (2002). Wahl M, Goecke F, Labes A, Dobretsov S, Weinberger F. Diversity and distribution of coral-associated bacteria. (2012). The second skin: ecological role of epibiotic Mar Ecol Prog Ser 243: 1–10. biofilms on marine organisms. Front Microbiol 3. Rosenberg E, Koren O, Reshef L, Efrony R, Zilber-Rosenberg I. Webster NS, Cobb RE, Soo R, Anthony SL, Battershill CN, (2007). The role of microorganisms in coral health, Whalan S et al. (2011). Bacterial community dynamics disease and evolution. Nat Rev Microbiol 5: 355–362. in the marine sponge rhopaloeides odorabile under Rosenberg E, Zilber-Rosenberg I. (2011). Symbiosis and in situ and ex situ cultivation. Mar Biotechnol 13: development: the hologenome concept. Birth Defects 296–304. Res C Embryo Today 93: 56–66. Whalan S, Webster NS, Negri AP. (2012). Crustose Schereiber U, Schiliwa U, Bilger W. (1966). Continuous coralline algae and a cnidarian neuropeptide trigger recording of photochemical and non-photochemical larval settlement in two coral reef sponges. PLoS One chlorophyll fluorescence quenching with a new 7: e30386. type of modulation fluorometer. Photosynth Res 10: Wilson S, Blake C, Berges JA, Maggs CA. (2004). Environ- 51–62. mental tolerances of free-living coralline algae (maerl): Schwarz AM, Hawes I, Andrew N, Mercer S, Cummings V, implications for European marine conservation Thrush S. (2005). Primary production potential Biol Conserv 120: 279–289.

Supplementary Information accompanies this paper on The ISME Journal website (http://www.nature.com/ismej)

The ISME Journal